Hyperspectral Image Unmixing Method and System Based on Abundance Significance Analysis

Abstract

The invention provides a hyperspectral image unmixing method and system based on abundance significance analysis, including establishing an endmember spectral library of a hyperspectral remote sensing image to be processed; using a mixed pixel decomposition method based on sparse regression to perform The preliminary mixed pixels are decomposed and arranged in descending order according to the abundance value, and the significance analysis is carried out on the sorted abundance sequence to obtain the critical value of the significance abundance, and then judge the composition according to the preset significance abundance threshold. The sparseness of the pixel represents the subset of endmembers; finally, the least squares method with abundance constraints is used to decompose the mixed pixel again, and the result is taken as the final result of the mixed pixel decomposition. The present invention can obtain a more sparse and accurate subset of pixel representation endmembers, and improve the resolution of mixed pixel decomposition of hyperspectral remote sensing images, in terms of high-precision ground object classification based on hyperspectral remote sensing images and ground target detection and recognition It has important application value.

Description

technical field [0001] The invention belongs to the technical field of remote sensing image processing, and relates to a hyperspectral remote sensing image mixed pixel decomposition method based on abundance significance analysis. Background technique [0002] Hyperspectral remote sensing images have high spectral resolution and can analyze the material composition of ground objects in a more detailed and precise manner. Therefore, they have attracted great attention from scholars at home and abroad and have been widely used in ground object classification, anomaly detection, and target recognition. . However, the spatial resolution of hyperspectral images is generally low, and a pixel usually contains the spectra of multiple ground objects, and these pixels are called mixed pixels. The ubiquity of mixed pixels greatly hinders the application of hyperspectral remote sensing images. The mixed pixel decomposition of hyperspectral remote sensing images is one of the core issu...

AI Technical Summary

Problems solved by technology

[0028] Although the mixed pixel decomposition method based on sparse regression emphasizes the sparsity of understanding, due to the interference of noise, the strong coherence of end-member spectra in the spectral library, and algorithm errors, the unmixed abundance is still far from the true value of the pixel. The abundance is sparse, and the number of endmembers in the representative endmember subset of the pixel is usually much larger than the actual number of endmembers contained in the pixel

Images